2014 Playoffs: (1) San Antonio vs. (2) Oklahoma City

benji's system is composed of columns of data that have no labels in a spreadsheet.
One of the fun things about it is it has ratings same as those found in nba live games.

This possession talk gave me deja vu, something about Avery Bradley.

Carry on nerds.

It's how he used to make his rosters, right?

The ratings are one of the legacy things I've never cleaned out. And I've still updated and tweaked them to where they're a hundred times better than they were when I did the Live PC thing because I don't know why, I'm an idiot.

Live's 20 ratings always seemed limited (especially after ID came out) but it's kinda hilarious to look at how they come out for certain players sometimes now. Especially the dumb way the FG rating operates.

And there's labels. There's just no explanation for them.

Actually, there are some descriptions, forgot I made these: https://docs.google.com/spreadsheet/ccc ... sp=sharing

They're for an outdated version of the sheet, almost as outdated as the Celtics sheet for this year I recently uploaded since after it I cleaned out a bunch of garbage*: https://docs.google.com/spreadsheets/d/ ... sp=sharing

*Even more so than I do when I post these sheets, since I clean out lots of odd little test things and the Live ratings usually.

Those Live ratings are the Joel Anthony of your sheets, you're going to regret removing those like the Heat will regret trading him away this season.

They're still in the actual file, I just usually delete them when I post the selected sheets.

Does benji have blog or something? Or even thread where he makes up or at least try to explain his system (or even his Live rating)? WTF is benji system?

Unless you accept the assumption that better possession distribution raises all boats. Kemba Walker, Gerald Henderson, etc. are all having better possessions because Jefferson is there using a bunch of possessions now.

And that's the point of using the Bobcats preview as an example, of course no predictions are entirely perfectly accurate, but WP doesn't even consider the effects of possession distribution. It simply says play the highest WP48 players at each position more. The Jefferson example, like Harden a year before shows the impact of adding a high usage player and if they're efficient (like Harden and Jefferson) so much the better.

Another example? Monta Ellis this year in Dallas. They're negative on both him and Carter when those were two of Dallas' key players this year. Considering the ESPN article about the Mavs offense, I think I'll lean towards a model that is in more agreement.

Just to use the players on the same team, if you give all of Carter's minutes to Jae Crowder and swap Ellis and Ellingtons, WP thinks the Mavs would be 53-29 instead of 49-33. My system thinks they'd would drop to 35-47 even with Dirk and Calderon boosting their individual ppg.*

Even more fun. Eliminating Dirk from this altered team and giving all his minutes to Dalembert, Wright and Blair.

Wins Produced:

Code:

Projection: 57-25
NAME   MPG   WINS
Jae Crowder   39.4   6.99
Wayne Ellington   36.9   5.98
Jose Calderon   30.1   7.51
DeJuan Blair   29.8   7.43
Shawn Marion   29.4   8.28
Brandan Wright   25.3   12.46
Samuel Dalembert   24.6   8.06
Devin Harris   10.0   1.30
Shane Larkin   6.0   -0.51
Monta Ellis   4.8   0.56
Gal Mekel   3.6   -0.96
Bernard James   1.8   0.12
Ricky Ledo   0.4   -0.07

Benjishit:

Code:

Projection: 25-57
Player   MPG   W
crowder,jae   39.3   3.77
ellington,wayne   36.9   3.55
calderon,jose   30.1   3.37
blair,dejuan   29.8   3.76
marion,shawn   29.4   2.95
wright,brandan   25.2   3.48
dalembert,samuel   24.6   2.72
harris,devin   10.0   0.82
larkin,shane   6.0   0.32
ellis,monta   4.8   0.35
mekel,gal   3.6   0.09
james,bernard   1.8   0.18
ledo,ricky   0.4   0.01

For funsies, their wins per 48 minutes and mine, sorted by mine:

Code:

Brandan Wright   .492   .138
DeJuan Blair   .249   .126
Jose Calderon   .249   .112
Samuel Dalembert   .328   .110
Bernard James   .067   .103
Shawn Marion   .282   .100
Wayne Ellington   .162   .096
Jae Crowder   .178   .096
Devin Harris   .130   .082
Monta Ellis   .118   .072
Shane Larkin   -.085   .054
Gal Mekel   -.270   .024
Ricky Ledo   -.185   .017

Very interesante.

Key piece? Vince Carter played lots of minutes for the Mavericks but he was just average. Note that his 15.9 PER came with 23.1 usage rate. His WS is slight over average too. Jae Crowder however was a low productive yet low usage player. I wouldn't be surprised if WP shows differently to WS in their comparison, especially when WP is position based. And this actually aligns with what Simple Rating(SRS) tells us as it's a metric that's opponent based. SRS also has Jae as better of the two. While it may not be entirely correct, SRS is a tool that tells us different things from other stats. So perhaps I am thinking WP works in a similar way as well.

b-r metrics are still my favorites but I don't find WP entirely phony business in many ways. Maybe it's hugely because I find their articles often refreshing and informative. It offers a different perspective.

Too many numbers. Hurts my brain

Nothing wrong with that, besides it's worse if you know such numbers and yet use them the wrong way just 'cause.

Like me!